Rec. West. Aust. Mus. 1988,14(1): 73-89

A new freshwater catfish (Pisces: Ariidae)from northern Australia

PatriciaJ. Kailola* and Bryan E. Pierce*

Abstract

A new species of fork-tailed catfish is described on the basis of 31 specimenscollected in northern Australia between the Fitzroy River (Western Australia) andthe Mitchell River (Queensland). Arius midgleyi sp. novo grows to at least 1.3 m TLand is distinguished from other Australo-Papuan ariids by a combination of charac­ters including snout shape, barbel length, eye size, tooth arrangement and gill rakernumber and position. Comparison is made with other ariid species occurring innorthern Australian rivers, including the morphologically similar A. leptaspis(Bleeker).

Introduction

The Timor Sea and Gulf of Carpentaria drainage systems (Lake 1971) approx­imately represent the Leichhardtian zoogeographic region of Whitley (1947). Therainfall pattern in this region is dominated by the wet monsoon (occurring withinthe period November to April). Most rivers here traverse a flat coastal plain about15 km wide before reaching the sea (Lake 1971). These rivers commonly possesswide flood plains and low gradients, often contracting to a chain of waterholesduring the dry season; some (Gregory River; Fitzroy to Daly Rivers) have reachesof rapids or very deep gorges. The average annual discharge from this region is69000 billion litres (Lake 1971), most of it occurring during the wet season.

Five of Australia's 18 species of fork-tailed catfishes (Ariidae) are common inthis northern region, yet were overlooked by Whitley (1947) and Iredale andWhitley (1938). The members of this family, which is distributed circumgloballyin the tropics and subtropics, may inhabit the sea, rivers within tidal influence, orfresh waters. Three ariid species present in these northern rivers (Cinetodus froggatti[Ramsay and Ogilby, 1886], Arius graeffei Kner and Steindachner, 1886, A.leptaspis [Bleeker, 1862]) occur in fresh and adjacent coastal waters and com­parative habitats in Papua. Like the relatively small A. berneyi Whitley, 1941,which only rarely penetrates brackish waters, A. midgleyi sp. novo is also a fresh­water fish. Although exploited commercially since 1978, the distinct taxonomicstatus of this species has been recognised only recently. The purpose of this paperis to describe this fish, to compare it with the other freshwater fork-tailed catfishesof northern Australia, and to present information on its biology.

* Department of Zoology, University of Adelaide, GPO Box 498, Adelaide, South Australia5001.

73

A new species of freshwater catfish

Materials and methods

Counts and measurements follow Hubbs and Litgler (1958) with the addition of:premaxillary tooth band length and width; 'interdorsal' fin space; occipital processbreadth and length; maxillary barbel length; free vertebral count (for explanation,see Kailola 1983). Counts of fin elements and gill rakers were made using a needleprobe; measurements were made with pointed calipers to the nearest 0.1 mm and,for very large specimens, the length was obtained using a mm-graduated ruler.Vertebral counts (number posterior to the anterior fused complex of 6 or 7vertebrae, and including the hypural) were made from clf;ared and stained spe­cimens and radiographs.

The range of counts and proportional measurements for the paratypes of thenew species are indicated in parentheses following the data for the holotype, ifdifferent.

In order to ascertain which factors contributed most to an explanation of thetotal variability, all variables were initially evaluated for specimens of both A.leptaspz"s and the presumed new species using principal component analysis(BMDP software package, subprogramme 4R; Dixon 1985). A reduced characterset including only the statistically significant variables, was used with clusteranalysis (BMDP, subprogramme KM; Dixon 1985) to substantiate the presence oftwo distinct groups - i.e. A. leptaspz"s and the new species. Interpretation of theseresults favoured acceptance of a two-groups hypothesis. The resulting populationsets (52 specimens of A. leptaspz"s and 28 of Arz"us new species) were then charac­terised within a stepwise discriminant function analysis (BMDP, 7M; Dixon 1985).Specimens (n=35) of A. leptaspz"s from New Guinea were later entered as 'un­knowns'to evaluate the accuracy of this classification function.

The following abbreviations are used in the text and tables: SL - standardlength; FL - fork length; TL - total length; HL - head length; D - dorsal fin;A - anal fin; P - pectoral fin; V - ventral (=pelvic) fin; C - caudal fin; GR - gillrakers; 1. - length; occip. - occipital; premax. - premaxillary; dist. - distance;interorb. - interorbital; w. - width; caud. - caudal; stn - collecting station.

Material has been deposited in the following institutions: the American Museumof Natural History, New York (AMNH); the Australian Museum, Sydney (AMS);the Northern Territory Museum, Darwin (NTM); the Queensland Museum, Bris­bane (QM); the United States National Museum, Washington (USNM); the WesternAustralian Museum, Perth (WAM); the Zoologische Museum, University of Amster­dam (ZMA).

Systematics

The new species is referable to the genus Arz"us as proposed by Valenciennes(1840). Taylor's (1986) diagnosis of Arz"us is appropriate for A. mz"dgleyz" sp. novoe.xcept for the presence of rakers along the rear margin of the first two gill arches.

74

I

I1

Patricia J. Kallola and Bryan E Pierce

However, based on examination of the type (A. arius Hamilton-Buchanan,' 1822)by one of us (PJK) we consider this character is of specific importance only.Because Arius contains numerous species, its eventual division into subgenera(based on a global comparison of type material) would seem the appropriatemethod by which the diversity of its species could be recognised.

The characters which contributed most to an understanding of the variability bet­ween the two taxa are: maxillary barbel length as a percentage of standard length,ratio of the length of the dorsal fin spine into head length, total gill raker counton the first arch, head width as a percentage of head le,.-lgth, and snout length as apercentage of head length (and see discussion under 'Remarks'). The resultingdiscriminant function [Zleptaspis =-366.336 + 12.523 (gill raker count) + 83.390(dorsal fin spine/HL) + 0.234 ([head width/HL] x 100) + 1.231 ([snout length/HL] x 100) + 0.0152 ([maxillary barbel length/SL] x 100); ZArius sp. novo =-369.429 + 9.722 (gill raker count) + 51.859 (dorsal fin spine /HL) + 0.874 (headwidth/HL x 100) + 1.081 ([snout length/HL] x 100) + 1.145 ([maxillary barbellength/SL] x 100) - where the highest Z score determines membership] separates100 per cent of the cases accurately in the robust jackknife classification matrix(BMDP, 7M; Dixon 1985). New Guinea specimens of 'A. leptaspis', which mightbe expected to differ somewhat from the Australian stock of this species, werecorrectly classed into that species group 100 per cent of the time using thisclassification scheme.

Based on this clear morphological distinction witnessed over many years ofsampling for catfish of both sexes from a broad geographic range, we concludethat A. midgleyi sp. novo is reproductively isolated from other Australian ariidspecies with which it is sympatric (particularly A. leptaspis and A. graeffei).Therefore, our material meets the criteria for designation as a species (Mayr 1963).

Arius midgleyi sp. novo

Figures 1-8; Table 1,2

Arius leptaspis - Taylor, 1964: 81 (in part - see 'Remarks')Arius species 1 - AlIen, 1982: 30.Arius species 1 - AlIen, 1982: 31.

HolotypeAMS 1.20858-006, 270 mm SL, Wickham Gorge, Victoria River, Northern Territory, 7 June

1978, collected by D.F. Hoese.Condition of holotype: Good. Short slit along belly; membrane on all fins split; tip of upper

caudal lobe lost.

Paratypes

Western AustraliaWAM P.25597-001, (1 specimen), 348 mm SL, Fitzroy River, June 1973, collected by R.

Emiliani; WAM P.25708-001, (1), 224 mm SL, Forbes Yard, Traine River, 18 June 1973,

75

A new species of freshwater catfish

collected by J.B. Hutchins; ZMA 119.467, (1), 244 mm SL, near Wyndham in fresh waterhole, 12 April 1981, collected by L. Turner; WAM P.28776-001, (1), 166.5 mm SL (clearedand single-stained), Lake Argyle, mid-1980, collected by N. Morrissy; WAM P. 21338-002(previously included in WAM P.21338-001), (4), 133, 138, 149, 161 mm SL, Ord River, 4October 1971, collected by RJ. McKay and J. Dell; AMNH 57082, (1), 152 mm SL, OrdRiver below Duncan Highway crossing, 7 May 1969, collected by J. Nelson, D. Rosen andH. Butler.

Northern TerritoryAMNH 57454 (previously included in AMNH 51649), (1), 98.5 mm SL, junction of Big

Horse Creek and Victoria River, 8 May 1969, collected by J. Nelson, D. Rosen and H. Butler;AMNH 57454SW (previously included in AMNH 51649), (3), 103, 106 and 114 mm SL, samedata as AMNH 57454; NTM S. 11800-001, (1), 325 mm SL, Daly River on Florina Station,25-26 August 1980, collected by H. and M. Midgley; QM I.16735, (1), 240 mm SL, HodgsonRiver, 17 September 1979, collected by H. and M. Midgley; QM I.16737, (1), 310 mm SL,Mannaburoo Hole, Limmen Bight River, 1 September 1979, collected by H. and M. Midgley;QM 1.16738, (2) (one broken), 327 mm SL and 151.5 mm HL, Mannaburoo Hole, LimmenBight River, 29-30 August 1979, collected by H. and M. Midgley; NTM S. 12083-001, (1),331 mm SL, Wilton River, 25-27 September 1979, collected by H. and'M. Midgley; NTM S.12070-001, (2), 298 and 315 mm SL, Mannaburoo Hole, 29-30 August 1979, collected by H.and M. Midgley; NTM/AS F.35, (1), 257 mm SL, Wollogorang Homestead, 15 June 1974,collected by D. Howe; NTM/AS F.36, (1),273 mm SL, same data.

Queensland.QM 1.12910, (1),326 mm SL, Flinders River near Maxwellton, 14 October 1974, collected

by H. and M. Midgley; QM 1.12757, (1), 310 mm SL, same data; QM 1.16730, (2),329 and315 mm SL, Flinders River in Maxwellton area, October 1974, collected by H. and M. Midgley;QM 1.11364, (1), 205 mm SL, Forest Home Station, Gilbert River, 24 September 1953, collec­ted by T.C. Marshall; AMS 1B.2882, (1),171 mm SL, same data; QM I.11990, (1),145.5 mmSL, Mitchell River, 8 September 1959, collector not 'stated.

DiagnosisA sleek catfish, attaining a large maximum size (known to 1.3 m). Barbels thin

and short, rarely reaching beyond pectoral fin base and less than 25 per cent SL;jaws strong, upturned slightly at symphyses, mouth broad; snout truncate inprofile; head oblong, its width averaging 66 per cent HL. Occipital process narrow,with parallel borders. Numerous fine, sharp teeth on palate in transverse band offour oblong groups. No rakers on posterior aspect of gill arches. Gill rakers onfirst arch few, 10-17; A 16-19. Number of free vertebrae 47-50.

A. midgleyi is distinguished from A. graeffei and A. berneyi by lacking rakerson the posterior (inner) aspect of the first and second gill arches (present on allarches in these two species). From Cinetodus froggatti, it differs in having a muchbroader mouth, four groups of teeth in a band across the palate (two patches onlyin C. froggatti) and wide gill openings extending well forward from the isthmus(restricted and terminating opposite lower pectoral base in C. froggatti). A. lep­taspis is most similar to A. midgleyi. The two species may best be separated onthe relative length of the maxillary barbels (Figure 7) (16-25 per cent SL, cf22-51 per cent SL in A. leptaspis). The maxillary barbel rarely reaches as farposteriorly as the head margin in A. midgleyi, but reaches and extends beyond

76

Patricia J. KaUola and Bryan E. Pierce

the pectoral spine base and further in A. leptaspis. Additional characteristicswhich distinguish these two species are: the combination of head width (56-76per cent HL, cf 66-83 per cent HL in A. leptaspis) (Figures 5, 6, 8), snout shapeand the total number of rakers along the face of the first gill arch (10-17, cf13-22 in A. leptaspis) (Table 1). The shape of the occipital process and extent ofthe granular head shield also differs between the two species, A. leptaspis havinga broader process and more extensive shield, the dorsomedian head groove ter­minating a considerable distance before the base of the occipital process. More­over, A. leptaspis is only known to a maximum body size of 60 cm FL (Bishopet al. 1986).

DescriptionThe range, mean and standard deviation of meristics and selected proportions

are given in Tables 1 and 2.D 1,7. A 18 (16-19). P I,ll (9-11). V 6. C primary rays 7 +8. GR (first arch

16 (10-17) of which 6 (3-6) on upper limb. GR (last arch) 18 (11-19). Numberof free vertebrae - (47-50; 17 specimens, mean 48.3). Branchiostegals 6.

Table 1 Meristics and relative body proportions of Arius midgleyi. (Ratios could not becomputed if character is damaged or missing on a specimen.) n=sample size; SD=standard deviation.

Character Holotype Paratypes

n range mean SD

SL (mm) 270 29 98.5-348 233.7 83.4HL in SL 3.1 29 3.0- 3.5 3.2 0.1eye 1. in HL 7.5 30 4.6- 11.4 7.8 2.1eye 1. in snout 1. 2.7 29 1.6- 4.2 2.8 0.8eye 1. in bony inter-

orbital width 2.5 28 1.3- 4.1 2.6 0.9occip. process width

in process 1. 2.8 29 1.5- 2.8 2.0 0.3D. spine 1. in HL 1.5 28 1.4- 2.4 1.8 0.3P. spine 1. in HL 1.5 26 1.6- 2.4 1.9 0.2adipose fin base in

D. fin base 1. 1.1 28 2.1- 3.7 2.8 0.4caudal peduncle depth

in its 1. 2.2 29 1.7- 2.6 2.1 0.2predorsal 1. in SL 2.4 29 2.4- 2.7 2.5 0.1longest barbel in SL 4.6 29 4.1- 6.0 5.1 0.5head height in head width 1.4 29 1.2- 1.7 1.4 0.21. premax. tooth band

in its width 9.5 28 5.2- 9.9 7.7 1.3count of A. fin rays 18 29 16 - 19 17.6 0.9count of P. fin rays 11 29 9 -11 10.4 0.6total GR (first arch) 16 28 10 - 17 13.1 26total GR (last arch) 18 25 11 19 15 2.4

77

Table 2

A new species of freshwater catfish

Percent of HL and SL for Arius midgleyi.

Character Holotype Paratypes

n range mean SD

Percent of HLhead height 46 29 40-56 47.5 4.6head width 67 29 56-76.5 66 3.9eye 1. 13 30 9-22 14 3.9mouth gape 49 24 41-51 46.5 3.3internostril dist. 35 24 29-39 33.5 3.3snout 1. 36.5 29 33-40 36 1.8longest barbell. 67 30 50-76 63 7.4bony interorb. dist. 33 28 26-38 33 3.4occip. process 1. 32 30 25-35 30 2.4occip. process w. 11 29 11-19 15 2.1

Percent of SLHL 32 29 29-34 31 1.3head height 15 28 13-17 15 1.1head width 21 28 18-24 21 1.5eye 1. 4 29 3-6 4 1.1mouth gape 16 23 12-17 15 1.5internostril dist. 11 23 9-12 10.5 1.3snout 1. 12 28 10-13 11 0.7longest barbell. 22 29 17-24 20 2.1bony interorb. dist. 11 27 8-12 10 1.3occip. process 1. 10 29 8-11 9 0.7predorsal 1. 41 29 37-42.5 40 1.4D. fin base 1. 10 29 9-12 10 0.8interdorsal space 26 28 23-29.5 26 1.6adipose fin base 1. 9 28 7-11 9.5 1.0A. fin base 1. 15 29 12-15.5 14 0.8caud. peduncle depth 7 29 6-7 7 0.3caud. peduncle 1. 15 29 12-18 14.5 1.1P. fin spine 1. 19 25 13-19.5 17 1.6D. fin spine 1. 21 27 13-21 18 2.5

Body robust, rather compressed, tapered posteriorly (Figures 1, 2). Lateralhead profile triangular and narrow. Predorsal profile straight, interorbital flat.Snout almost truncate, broad, projecting short distance beyond lower jaw inyoung; mouth subterminal in adults. Lips rubbery, 'thin at front of jaws and ratherthick at corners; jaws strong, slightly elevated at symphysis - particularly thelower. Mouth broad, curved; band of teeth in upper jaw not or partly visible whenmouth closed, though often just visible at sides of mouth. Nostrils ovate, placedwell forward; anterior nostril directly before or slightly lateral to posterior one onwhich skin flap just conceals opening. Short crescentic groove on snout of smallspecimens. Eye rounded to almost oblong, orbit free from head skin; eye placeddorsolaterally and visible when head viewed from above; mid head length at or up

78

Patrioia J. Kallola and Dryan E. Pieroe

to one eye diameter's distance, behind hind margin of eye; prominence of lateralethmoid sometimes apparent in larger fish. Gill openings broad, membranesmeeting at sharp angle well forward, margins broadly free of isthmus.

Figure 1 Drawing of the holotype of Arius midgleyi sp. nov., AMS 1.20858-006. 270 mm SL.

----Figure 2 Arius midgleyi paratype. WAM P.213380-001, 161 mm SL.

~~~~

'~

Figure 3 Semi-diagrammatic view of upper tooth pattern in A. midgleyi (ZMA 119.467),where pmx=premaxillary tooth band, vo=vomer tooth patches, ect='ectopterygoid'tooth patches.

79

A new species of freshwater catfish

Numerous small, sharp and depressible teeth arranged in band of irregular seriesin jaws, embedded in fleshy tissue: 16 to about 24 in upper jaw, 10-15 in lowerjaw. Lower jaw band divided at symphysis by narrow edentulous space. Teethpresent in four oval patches on palate (Figure 3), transversely arranged: patchesalways distinct and not contiguous, inner pair more rounded and about two-thirdswidth of outer pair; 8-12 series of teeth on smaller patches, 12-18 on larger. Palatesmooth with sometimes a low, diagonal ridge of skin on each side posteriorly.Skin lining branchial chamber loose towards margin, and forming a deep pocketwith narrow opening.

Head shield usually concealed by skin and mucus in juveniles and often in fresh,larger fish. When exposed, shield very granular, granules extending forward toabove eye, to gill opening origin and over occipital process; predorsal plate rugoseor granular. Dorsomedian head groove long and lanceolate, and distinct - begin­ning at level of nostrils and tapering posteriorly to or almost to base of occipitalprocess, distal end bordered by loop of granular striae. Occipital process oblongand sometimes with low median keel, its sides straight. Head vcnulose overopercles. Numerous very fine papillae scattered over snout, anterior two-thirdsof head and sometimes breast on many specimens. Triangular humeral processsmooth to rugose, moderately ossified anteroventrally, its shaft oblique and acute,reaching two-thirds distance along pectoral spine. Axillary pore tiny, closelyadjacent to process.

Barbels thin and tapered. Maxillary barbel reaching preopercular margin toalmost head margin above pectoral base; mandibulary barbel reaching or fallingwell short of, ventral head margin; mental barbel extending to below middle orhind border of eye.

Gill rakers rigid, strong and rather sharp-tipped, as long as gill filaments. Rakersabsent from posterior face of first and usually second arches; 11-17 (mean 15.2)rakers along rear of third arch. A low, thick pad of tissue usually present postero­dorsally on second arch, other arches lacking such thickening.

Fin spines sharp, moderately to very thick, compressed. Fine longitudinal striaelaterally on spines, outer (anterior) border roughened by granules and low dentae,few antrorse serrae near tip; inner (posterior) border of dorsal with no or fewserrae extending half-way from apex; usually up to 20 saw-like, short serrae alonginner border of pectoral spine. Tip of spines with short filaments. Dorsal spinelonger than pectoral: longer than postorbital in young, about two-thirds in adults.Dorsal fin rather high, longest ray 2.7 (2-3.5, mean 2.9) times last ray. Pectoral finreaching to below posterior dorsal rays. Adipose fin oblong and high, placedopposite middle of anal fin. Longest anal ray 3 (1.9-3.5, mean 3.1) times length oflast ray; fin moderately elevated anteriorly, with concave outer margin. Caudalfin lobes long, broad basally, slender and tapered distally. Ventral fin moderatelybroad in females, reaching to anal fin origin or up to fourth anal ray, maturingfish with thick pads on fifth and sixth (inner) rays; in males and juveniles, finsnarrow, never reaching and often falling quite short of, anal fin.

80

Patricia J. Kailola and Bryan E. Pierce

Lateral line sloping over anterior half of body, thence straight, slightly elevatedon caudal fin base. Distinct vertical series of pores extending from line over backand lower sides; short oblique venules diverging along line's length, numerous andlonger over 'shoulder'. Caudal peduncle moderately deep and compressed.

Colour when fresh: highly variable, perhaps relative to habitat and locality.Specimens from the Katherine River ochre-coloured, brownish above, creamybelow; fins pale brownish orange, barbels pale. Ord and RopeI' River individualscountershaded olive brown, 'smokey' blue to dark blue above, white below a linefrom upper jaw along mid-sides to above anal fin or lower caudal peduncle; dorsal,adipose and caudal fins brown to dark bluish brown, pectoral and ventral finsdark (blue) above, cream below; anal fin brown or bluish brown, anterior andposterior of fin cream, sometimes only posterior rays dusky cream; maxillarybarbel blue, others white. Some individuals with 'piebald' colouration have alsobeen taken from Lake Argyle and Flora River: body greyish brown with small andlarge irregular cream or black blotches over anterior two-thirds of body especiallyaround mouth and on head. Peritoneum pale grey or pinkish. Fat bodies deepyellow or orange.

Colour when preserved: varies from brown above and fawn below to blackishor dark charcoal-blue above, fawn, orangey, cream or white below or blotched.Fins and barbels as above but blue becomes dark brown.

Distribution (Figure 4)Common and widespread in rivers and associated fresh waters of north-western

and northern Australia from the Fitzroy River through the Ord and VictoriaRiver systems (including Armstrong, Camfield, Humbert, Wickham, East Baines,West Baines Rivers, Neave and Waterloo Creeks Midgley 1981), the Keep River(rare) (Midgley 1981), Daly River, Katherine River, Flora, Fergusso!l, Fish andDouglas Rivers (Midgley 1980), McKinlay, Mary, South Alligator, East AlligatorRivers, RopeI' River system (Limmen Bight, Roper, Hodgson, Wilton and MainoruRivers - Midgley 1979, 1983), McArthur River, Tooganginie Creek (Midgley1983), Robertson and Calvert River systems (Midgley 1982) (rare), south-easterly(rare) to the Flinders River (near Maxwelton), Gilbert, Staaten Rivers and EdwardRiver system (Strathgorden Lagoon) on Cape York Peninsula (Midgley, pers.comm. 1986). (The record of distribution in the type series is supplemented herefrom specimens collected by H. and M. Midgley - the identification of a rep­resentative sample of which we have verified - and from additional material[non-type] we have determined and list herein.)

The four additional ariid species are distributed in northern Austalia as follows:Arius graeffei is sympatric with A. midgleyi throughout the range of the latter;C. froggatti is restricted to the RopeI' River drainage; A. berneyi extends through­out Queensland, but no further west than the RopeI' River system; and A. lep­taspis extends from northern Queensland (west or north-draining rivers) just intothe western Northern Territory.

81

A new species of freshwater catfish

tFitzroy R,

Figure 4 Australia north of 22°S, showing the distribution of Arius midgley; • representsmaterial determined by us; 0 represents additional records by H. and M. Midgley.

EtymologyThis large catfish is given the surname of Hamar and Mary Midgley who first

brought its existence as a distinct taxon to the attention of the first author andin part recognition of their enthusiastic and dedicated study of the fresh watersof northern Australia.

Habitat notes

Fresh water, only a few records from water near the upper limit of tidal influence.Found in fast-flowing main rivers, billabongs, creeks, deep pools and drying-outwaterholes. Water conditions vary considerably from turbid (10 cm visibility) tovery clear (4 m visibility), still or moderately-fast flowing. The water is alkaline(ph 7-8.7) and warm (surface temperature 22.5-35°C) (AlIen 1982; Midgley1979-83). A. midgleyi is possibly replaced in downstream reaches by one or bothof the other common ariids in northern Australia: A. leptaspis and A. graeffei.For example, Bishop et al. (1986) caught no A. midgleyi during their extensivesurvey of the ]abiru-Alligator Rivers area; and neither did Midgley (1984). Al­though A. leptaspis is absent from Lake Argyle, A. graeffei makes up about 30per cent of the fork-tailed catfish population in the impoundment.

Biological notes

In establishing the validity of A. midgleyi as a taxon, we collected information onthe species' biology which further characterises it as distinct from sympatricariids. Independent observations, together with our own, substantiate the unbiasednature of our conclusions.

Maximum size: to about 28 kg (S. Sharpe, pers comm. 1987); to approximately1.3 m TL (R.]. McKay, pers. comm. 1978; S. Sharpe, pers. comm.).

82

Patricla J. Kallola and Bryan E. Pierce

Diet: primarily predatory, but becoming a facultative omnivore during the dryseason. Recorded gut contents are fish (notably the bony bream, Nematalosaerebi (Gunther, 1868), in Lake Argyle (N. Morrissy, pers. comm.), aquatic andterrestrial insects and insect larvae, beetles and crustacean fragments (probablyMacrobrachium sp.). Such items were present in the stomachs of specimensprepared for osteological study by one of us (PJK). The size of food items waslarge in large fish. In a study conducted between August and September, Midgley(1981) recorded prawns, aquatic beetles, terrestrial beetles and grasshoppers inthe stomachs of A. midgleyi specimens.

Breeding: A. midgleyi breeds only in fresh water with no recorded spawningmigrations. Midgley (1979, pers. comm. 1986) has caught juveniles as small as63 mm SL from riverine situations (e.g. upper reaches of the Flinders River) incompany of larger adult fish. We have not examined fully mature fish, and littleinformation is available, probably because surveys have often been conductedduring the late dry season. Specimens captured in August from the Roper Riverarea are in the early stages of maturation suggesting that A. midgleyi conformsto the general early wet season spawning pattern of A. graeffei and other ariids(Rimmer 1985). In the Victoria and Daly rivers areas, individuals at gonadalmaturity stages III or early IV (sensu Snyder 1983) were obtained during Sep­tember and October and were more mature than the sympatric A. graeffei pop­ulation (Midgley, pers. comm.). Lake Argyle individuals captured during Sep­tember are in maturity stage V (S. Sharpe, pers. comm.). From 35 fish netted inLake Argyle at the end of July 1980, N. Morrissy (pers. comm.) determined 20to be maturing females, stages 11 to IV. The remaining 15, of the same size range,were either immature or non-ripening fish (both sexes). Morrissy noted difficultyin ascertaining the sexes of immature and 'resting' fish, a problem recognised byK. Bishop (pers. comm.) and Rimmer (1985). In Morrissy's sample of 20 maturefish, 13 had ova of 1 cm diameter (52 to 87 cm FL), five had ova of 0.5 cmdiameter (58 to 76 cm FL) and two had ova of less than 0.2 cm diameter (53 to57 cm FL).

Morrissy (pers. comm.) noted that females between 70 and 80 cm FL haveabout 100 ova in each gonad. K. Lightburn and B. Host (pers. comm. 1980)counted up to 180 ova (total) in mature females from Lake Argy1e, and S. Sharpe(pers. comm. 1987) estimated a single spawning fecundity of 100 to 400 ovaover the size range of mature fish. None of our informants ascertained whetherthe count differed between gonads.

Subsequent work on A. midgleyi in Lake Argyle (N. Morrissy, pers. comm.)employed length frequency analysis to determine that females matured at thebeginning of their third year of life at approximately 50 cm FL.

Growth and behaviour: in Lake Argyle, where the population has been fishedcommercially since 1978 (Morrissy 1983), A. midgleyi has been observed movingin large schools of similar size individuals. The brooding males congregate in

83

A new sPecies of freshwater catfish

Figure 5 Comparison of head shape of A. midgleyi (RHS, holotype) and A. leptaspis (LHS,QM unreg., Gulf of Papua, 235 mm SL) when viewed from above (preserved spe­cimens).

Figure 6 Comparison of head shape of A. midgleyi (LHS, NTM S.11153-001, Mainoru R.,NT, 110 mm HL) and A. leptaspis (RHS, NTM S.11153-002, same locality, 113 mmHL) when viewed from below (preserved specimens).

84

Patticia .1. Kailola and Bryan E Pierce

deeper water in a manner comparable to that reported by Aldaba (1931) anMane (1929) of Laguna de Bay (Philippines) ariid catfishes.

A length-frequency study on 650 individuals was conducted by N. Morrissin Lake Argyle, between 30 July and 7 August 1980. The fish were gill nette.(2.5-18 cm mesh) in about 10 m depth. Morrissy caught only three fish less that20 cm FL (probably the 0+ year class*), most fish were 25 to 50 cm FL witla total body weight to 2 kg (presumably the 1+ year class), and the remaindelbetween 55 and 75 cm FL (presumably the 2+ year class). S. Sharpe, a LakeArgyle commercial fisherman, informed us (1987) that A. midgleyi juvenilesattain a length of 20 to 30 cm within their first year. A. midgleyi in either case,exhibits a very rapid growth rate under these conditions relative to other northernAustralian ariids. It is faster growing than A. graeffei in this reservoir and indeed,is the largest truly freshwater fork-tailed catfish in Australia. £1. midgleyi isfavoured by conditions in this impoundment, as it is seldom numerically dominantelsewhere in its range.

RemarksA. midgleyi is immediately recognisable in the field because of its rather oblong,

'square' snout (Figures 5, 6) and broad mouth (approximately equal to the out­side distance between the eyes ~ Midgley 1981). Dubbed 'squarenose' or 'shovel­nose' by fishermen, it looks somewhat shark-like underwater.

At our request, Janet Gomon (USNM) examined some of Taylor's (1964) spe­cimens of 'A. leptaspis' from northern Australia. Based on our diagnoses, shebelieves that Taylor's specimen from near the RopeI' River Mission (USNM 173564)belongs instead to A. midgleyi.

Allen and Hoese (1980) considered that 10, possibly 13 fish species present inthe eastern and western extremities of the Gulf of Carpentaria drainage systemexhibit disjunct distribution patterns; and in view of its rare occurrence in thesouthern Gulf of Carpentaria drainage, A. midgleyi appears to fall into thiscategory. We believe that such a phenomenon may equally be an artefact ofcollecting effort. In view of studies conducted in aquaria by one of us (BEP)on four of the species cited (individuals maintained at 16-20° C over winter, twospecies breeding at 20°C), it is our opinion that temperature may not be thesole cause of the apparent disjunct distribution.

Alien and Hoese (1980) conjectured that a disjunct distribution pattern may becaused by lower winter water temperatures in streams flowing into the southernGulf of Carpentaria: here the temperatures decrease to 22-23°C during winter,compared to 27-32° C in northern streams (Alien and Hoese 1980). Extrapolatingfrom Nix and Kalma (1972) (who postulated a lowering of 5-6°C in water tem­perature during the Pleistocene), these authors reasoned that minimum water

* (Supported by the similar sized specimens from AMNH 51649 which were eviscerated priorto clearing: SL's 103, 106, 114, they had abundant yolky-coloured fat in the body cavityand were captured early May: probably 0+ juveniles.)

85

Figure 7

Figure 8

A new species of freshwater catfish

600 0 00 00

0

0 0

J::0

0 0 0

-'" 00., c 500 0.0., 0 0 0

:0-' 00 00

lIlu 0

-:0 0 0"'u., c0"'",c_ o •

3(1) 400

00J::c-., ."'0c ~ . ..,., ·..-,a. 300 . .

'" . .'"

...200

100 200 300 400

Standard Length

Comparison of longest barbel length between A. midglcyi (0) and A. leptaspis(e) expressed as per cent of SL.

1800

J::

'" 170 0c., 0

-' 0

u

'"., 160:I: 000

'0 0

c000

00., 0

0 150 0 00 0

Q; 0

a. 0 0.. 00 0

'" 0 0

'" 0

· .J:: 140U .~u

'" 130 0 ..,:I: ·. . .. . ..... .· ...

120.

100 200 300 400

Standard Length

Comparison of head width between A. midgleyi (0) and A. leptaspis (e) expressedas per cent of SL.

86

Patrlcia J. Kailola and Bryan E. Pierce

temperatures to at least 18-19° C in the lower Gulf would have proved an effectivebarrier during that period to the southerly dispersal of some extant fishes. How­ever. if temperature were the key factor involved, it begs the question of whyisn't A. midgleyi known from southern New Guinea? From our understandingof Nix and Kalma (1972) and associated papers, we believe that changes in otherabiotic conditions (for example, geomorphology, aridity, seasonality, watersalinity and river flow) besides temperature, have likely contributed to the present­day distribution pattern of northern Australian freshwater fishes.

Additional non-type material examined

Western AustraliaUnregistered, (13), Lake Argyle, 1980, 170-385 mm SI.; AMS I.l6838-006, (2), Ord River,

1971,137 & 151 mm SI..

Northern TerritoryNTM S.12077-001, (1), Roper River, 1979, 107.5 mm HI. (cleared and stained); QM I.

16743, (1), Wilton River, 1979, upper jaw only, SI. not stated; QM I.l6744, (1), I.l6745,Mainoru River, 1979, upper jaw only, SI. not stated; QM I.l6745, (1), same data, upper jawonly, SI. not stated; NTM S.I1153-001, (1), Mainoru River, 1979, 110 mm HI.; AMNH 51650,(1), Katherine River, 8 mi. downstream from Katherine, 1969, 125 mm SI.; NTM S.12087-001,(3), Katherine River near Limestone Creek junction, 1980,355-385 mm SI. (111.6, 120.5,117.2 mm HL); AMS I.l6838-007, (3), same data, 127-151 mm SI.; AMS 1.20924-008, (1),Mary River, 1978,310 mm SI.; AMNH 51649, (2), junction of Big Horse Creek and VictoriaRiver, 1969, 104 & 122 mm SI.; AMS 1.20858-005, (8), Wickham Gorge, Victoria River, 1978,255-320 mm SI.; AMS 1.20848-010, (1), Jasper Gorge, Victoria River, 1978, 87 mm SI.; AMS1.20856-004, (2), BuBo River (Victoria River drainage), 1978, 212 & 225 mm SI.; AMS 1.20847­008, (1), upstream from Daly River crossing, 1978,240 mm SI.; AMS 1.20857-002, (1), Vic­toria River district, 1978,240 mm SI.; AMS 1.20453-001, (1), McKinlay River, 1978,250 mmSI.; AMNH 51651, (16), South Alligator River on Pine Creek·Oenpelli road crossing, 1969,118-224 mm SI.; AMNH 51650, (1), Katherine River, 8 miles downstream from Katherine,1969, 125 mm SI. QM I.l6744, (2), Mainoru River, 1979, upper jaws only, SI. not stated;QM I.l6745, (2), same data, upper jaws only, SI. not stated; QM I.l6743, (1), Hodgson River,1979, upper jaw only, SI. not stated.

QueenslandQM I.l6741, (2), McArthur River, 1975, 123.5 & 130.7 mm HI..

Acknowledgements

Our presentation of the biology of A. midgleyi has been greatly facilitated byStephen Sharpe (Silver Cobbler Fisheries, Kununurra, WA), Hamar Midgley(Nambour, Qld) and Noel Morrissy (WA Marine Research Laboratories, WA).The following persons kindly sent us material for study: N. Morrissy, H. and MaryMidgley, John R. Paxton and Douglass F. Hoese (AMS), Gerry R. Alien (WAM),Darryl Grey (NT Fisheries), Helen K. Larson (NTM) and M. Norma Feinberg(AMNH). We are indebted to Phil Kempster and Ruth Evans (Dept of Zoology,Adelaide University) for illustration assistance, to Ross McDonald (AdelaideDental School) for radiography, and to Michael J. Tyler (Dept of Zoology,

87

A new species of freshwater catfish

Adelaide University) and two anonymous reviewers for comments on the manu­script.

References

Aldaba, V.C. (1931). The Kanduli fishery of Laguna de Bay. Philipp. j. Sci. 45 (1): 29-39.Alien, G.R. (1982). A field guide to inland fishes of Western Australia. Western Australian

Museum, Perth, pp. 86.Alien, G.R. and D.F. Hoese (1980). A collection of fishes from the jardine River, Cape York

Peninsula, Australia. j. R. Soc. W. Aust. 63 (2): 53-61.Bishop, K.A., Alien, S.A., Pollard, D.A. and Cook, M.G. (1986). Ecological studies on the

freshwater fishes of the Alligator Rivers region, Northern Territory. Volume 1: Outline ofthe study, summary, conclusions and recommendations. Res. Rept Superv. Sci. AlligatorRivers Reg. 4: pp. 55.

Dixon, WJ. (1985). BMDP statistical software. WJ. Dixon (ed.). University of CaliforniaPress, Berkeley, California, pp. 733.

Hubbs, C.L. and Lagler, K.R. (1958). Fishes of the Great Lakes region. Bull. Cranbrook Inst.Sci. 26: pp. 213.

Iredale, T. and Whitley, G.P. (1938). The fluvifaunulae of Australia. S. Aust. Nat. 18 (4):64-68.

Kailola, PJ. (1983). Arius graeffei and Arius armiger: valid names for two common species ofAustralo-Papuan fork-tailed catfishes (Pisces, Ariidae). Trans R. Soc. S. Aust. 107 (3):187-196.

Lake, j.S. (1971). Freshwater fishes and rivers of Australia. Thomas Nelson (Aust.) Ltd.,Sydney, pp. 61, 96 col. pI.

Mane, A.M. (1929). A preliminary study of the life history and habits of kanduli (Arius spp.)in Laguna de Bay. Philipp. Agric. j. 18 (2): 81-115.

Mayr, E. (1963). Animal species and evolution. Belknap Press, Harvard, pp. 797.Midgley, S.H. (1979). The Roper River system, Limmen Bight River system, Rosie Creek

system in the Northern Territory. A biological resource study of fresh waters conductedduring August-September 1979. Rept. Fish. Div., Dept Prim. Product, N.T. Unpubl. pp. 52.

Midgley, S.H. (1980). The Daly River and its principle tributaries Katherine River, King River,Dry River, Flora River, Fergusson River, Fish River, Douglas River in the Northern Territory.A biological resources study of fresh waters conducted during August-September 1980.Rept. Fish. Div. Dept Prim. Product, N.T. Unpubl. pp. 49.

Midgley, S.H. (1981). The Victoria River and its tributaries Angalarri River and its tributarythe Ikymbon River, Armstrong River, Camfield River, Wickham River and its tributary theHumbert River, West Baines River and its tributary Waterloo Creek, East Baines River,Neave Creek and jasper Creek. The Fitzmaurice River. The Keep River and its tributaryChinaman Creek in the Northern Territory. A biological resource study of fresh watersconducted during August-September 1981. Rept. Fish. Div., Dept Prim. Product, N.T.Unpubl. pp. 69.

Midgley, S.H. (1982). The Lake Eyre drainage. The lames River and Lorne Creek. The inlanddrainage. The Playford River, Brunette Creek, Corella Creek, Creswell Creek and NewcastleCreek. The Gulf of Carpentaria drainage. The McArthur River, Wearyan River, FoelscheRiver, Robinson River, Calvert River and Nicholson River systems in the Northern Territory.A biological resource study of fresh waters conducted during August to October 1982. ReptN.T. Fish. Ind. Res. Devt Trust Fund. Unpubl. pp. 72.

Midgley, S.H. (1983). The rivers of Arnhem Land. The Phelp River, Rose River, Walker River,Koolatong River, Giddy River, Goyder River, Wilton River, Mainoru River, Cadell River,

88

Patricia J. Kailola and Bryan E. Pierce

Liverpool River and Goomadeer River, and the Towns River immediately to the south-eastof Arnhem Land in the Northern Territory. An account of a biological resource study offresh waters conducted during July to August 1983. A private study by Midgley and Midgley,Consultants, with assistance of the N.T. Fisheries Division. Unpubl. pp. 44.

Midgley, S.H. (1984). The Reynolds River, Finniss River, Adelaide River, McKinley River,Mary River, South Alligator River and Karns Creek in the Northern Territory. A biologicalresources study of fresh waters conducted during August, September 1984. Rept N.T. Fish.Ind. Res. Devt Trust Fund. Unpubl. pp. 50.

Morrissy, N. (1983). Potential fisheries capability of the Ord in the short, medium and longterms. West. Aust. Dept Fish. Wildl. Rept No. 59: pp. 24.

Nix, H.A. and Kalma, j.D. (1972). Climate as a dominant control in the biogeography ofnorthern Australia and New Guinea. In: D. Walker (ed.), Bridge and Barrier: the naturaland cultural history of Torres Strait. Australian National University, Canberra, pp. 437.

Rimmer, M. (1985). Reproductive cycle of the fork-tailed catfish Arius graeffei Kner andSteindachner (Pisces: Ariidae) from the Clarence River, New South Wales. Aust. f. Mar.Freshwat. Res. 1985, 36: 23-32.

Snyder, D.E. (1983). Fish eggs and larvae. In: L.A. Nielsen and D.L. johnson (eds), FisheriesTechniques. American Fisheries Society, Bethesda, Maryland, pp. 468.

Taylor, W.R. (1964). Fishes of Arnhem Land. In: Rec. Amer.-Aust. Sci. Exped. Arnhem Land,vol. 4, Zoology, pp. 45-307.

Taylor, W.R. (1986). Family 59: Ariidae. Seacatfishes. Seebabers. In: M.M. Smith and P.C.Heemstra (eds), Smith's Sea Fishes. Macmillan South Africa (Publishers) (Pty) Ltd, jo­hannesburg, pp. 1047.

Valenciennes, A. in Cuvier, A. and Valenciennes, A. (1840). Histoire naturelle des poissons,15. Paris: Pitois.

Whitley, G.P. (1947). The fluvifaunulae of Australia with particular reference to fresh waterfishes of Western Australia. W. Aust. Nat. 1 (3): 49-53.

Received 8 October 1987 Accepted 10 March 1988

89

Published 10 June 1988